Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Mycotoxins are toxic secondary metabolites produced by molds, i.e., metabolites not essential to the normal functioning of the cells. Molds are ubiquitous in nature and are universally found where environmental conditions are conducive to mold growth. Because molds are present in soil and plant debris, and are spread by wind currents, insects, and rain, they are frequently found in/on foods together with their associated mycotoxins (1).

The acute toxicity of mycotoxins has resulted in serious human health problems throughout recorded history (2). It has only been since the early 1960s, when the aflatoxins were found to be carcinogenic, that it was realized that some of these mold metabolites might have significant sub-acute and chronic toxicity for humans. The public health concerns resulting from the finding of mycotoxins, including metabolites in animal tissues resulting from transmission of mycotoxins present in animal feeds, and the observation of both acute and chronic effects in animals has prompted the research effort focusing on analytical methods development. Analysis for mycotoxins is essential to minimize the consumption of contaminated food and feed.

Method development and evaluation for mycotoxins is not a simple task. Determining the concentrations of toxins in grains at the ng/g or parts-per billion levels required for the most important mycotoxins is difficult. The approach generally followed consists of obtaining a relatively large primary sample representing a lot, reducing it in bulk and particle size to a manageable quantity, and finally performing the analysis on a small representative portion. Sampling commodities for mycotoxin contamination follows the U.S. Department of Agriculture (USDA) recommendations, which require collection of laboratory samples of at least 5–25 kg of nuts, corn, milo, and other grains (3).

All analytical methods for mycotoxins consist of four steps: sample preparation, extraction, cleanup or isolation, and measurement of the toxins. To prepare a representative test portion for analysis, the laboratory sample is ground and mixed, so that the concentration of toxin in the small final test portion is the same as that in the original laboratory sample collected. The test portion is extracted with various solvents. The extract is filtered and applied to a cleanup column or immunoassay device, or partitioned with appropriate solvents. This partially purified analyte is further separated and determined by liquid chromatography, capillary electrophoresis, or measured after an immu-nochemical reaction.

Numerous individual methods have been published for mycotoxin analysis. For aflatoxin alone more than 8000 papers have been published; hence a great deal of judgment is required for the selection of the optimum protocol of analysis. The analyst must use authentic toxin standards of known purity and select appropriate methods according to particular needs. The following criteria are considered in selecting a method: number of analyses, time required, location of analysis (laboratory or field), cost of equipment, safety, waste disposal, and the experience required of the analyst (4). Some of the published methods have been validated in international collaborative studies, and their precision and accuracy estimated for specific commodities according to internationally harmonized rules. AOAC International is one of the organizations that administer collaborative studies following the AOAC-International Union of Pure and Applied Chemistry (IUPAC) harmonized procedure (5). Such studies typically yield both precision (intralaboratory relative standard deviation, RSDr) and reproducibility (interlaboratory relative standard deviation, RSDR) data. By examination of a large body of data generated in such collaborative studies, Horwitz came to the surprising conclusion that one could predict with some confidence the RSDR to be expected based solely on the concentration of the analyte, and largely independent of the matrix, type of analyte or type of method (6). The RSDR, when plotted as a function of the concentration (expressed as a decimal fraction), described a curve, the so-called “Horwitz Horn” Fig. 1 ) (7). Thus at 1 μ g/g (C, expressed in toxin mass/commodity mass unit (g), 1 ppm C = 10−6) the expected precision would be approx 16%; at 1 ng/g (1 ppb) it would be approx 45%, etc. As a consequence, below 1 ng/g, when the RSDRs tend to rise above 50%, the results become uninterpretable because of the appearance of excessive numbers of false positive and false negative results. Such data are considered to be “out of statistical control.” Analyses can be conducted below 1 ng/g in specific laboratories if extraordinary steps are taken to ensure quality control.

Fig. 1.
figure 1

The “Horwitz Horn..”

Full size image

The now well-known “Horwitz Equation,” RSDR = 2C−0.1505 quantified the relationship between the RSDR and analyte concentration (8). C is in terms of a decimal fraction (ranging from pure compound, g/g, C = 1, to ultratrace toxins, ng/kg; C = 10−12). Hall and Selinger characterized this relationship as “one of the most intriguing relationships in modern analytical chemistry” (9). A heuristic derivation of this relationship was recently published based upon the observed change in standard deviation(s) as a function of concentration (8). It can easily be shown, using this relationship, that the RSDR will increase by a factor of two for each decrease in analyte concentration by two orders of magnitude.

The Horwitz Equation is extremely useful in evaluating analytical methods, i.e., their fitness for the intended use. A collaborative study is run, and the results are compared with those predicted by the Horwitz Equation. This is done by calculation of the Horwitz Ratio (HORRAT) (6): HORRAT = RSDR (found)/RSDR (predicted)

Based upon data generated on thousands of samples analyzed in a large number of collaborative studies (each study often included more than 8 collabora-tor-analysts of international laboratories) over many years, it was concluded that a HORRAT>2 indicates that the method is not acceptable, i.e., leads to poor results in the hands of experienced analysts in well run laboratories. Values bracketing 1.0 indicate acceptable precision and a method clearly under statistical control. Experience has shown that high HORRAT values usually indicate that (a) the method as written is problematic; (b) the samples distributed to the collaborators were nonhomogeneous; or (c) the laboratories involved had difficulty with the preparation, retention, or stability of standards. Low HORRAT values (<0.5) also have some significance, indicating either (a) lack of independence in the analyses; or (b) the use of advanced technology and strict adherence to a QA program by the laboratories involved. Lack of independence often results from (a) unreported consultation among participants; (b) replication by analysts until results are in agreement; (c) averaging by individual analyst of multiple analyses; or (d) excessive rounding.

The vast majority of published methods have not been validated by interlaboratory collaborative study. However, even with these, one can estimate an approximate HORRAT and come to some defensible conclusions relative to their capabilities.

In Table 1 are listed some of the mycotoxin methods which have been collaboratively studied and have been approved as “Official Methods” by AOAC International and, as a consequence, are often used by the US Federal and State regulatory laboratories (10). It gives commodities, levels, statistics, and HORRAT values. The HORRAT were obtained by comparing RSDR of the particular study to the RSDR of the corresponding concentration C from Fig. 1 In the process of validation, the method is first submitted to collaborative study (test samples with various toxin levels in duplicate are analyzed as unknowns by 8–15 laboratories), and the results are evaluated by a group of expert analysts (the safety officer, the general referee of the topic, the associate referee of the study, the statistician, and the method committee, and the AOACI Official Methods Board, OMB). If the study was properly conducted according to the “harmonized protocol,” if the analytical data fulfills the required statistical parameters, and if the reviewers’ comments (general referee, the statistician, members of the method committee) have been addressed, the OMB will accept the method as a “First Action AOACI Official Method.” After two years of successful use the method is adopted as final action and is published as an “Official Method” in the AOACI Compendium of Official Methods.

Table 1 Some of the AOACI Official Methods for Mycotoxins
Full size table

The use of a validated (Official) methods or published methods does not automatically ensure that a laboratory will produce acceptable results. It should be clearly understood that such methods are developed for use with a particular commodity (matrix). Should the method be used for other matrices, it is necessary for the analyst to evaluate applicability by conducting recovery studies with emphasis on the acceptability of the results at the level of toxicological concern. The analyst must compare results of his or her studies on the material in question with data of the collaborative studies of published methods. A laboratory quality assurance program should be implemented. Attention should be given to standard operating procedures, sample integrity and traceability, reference material, standard, control charting, and record keeping. The participation in proficiency testing programs is also recommended (11).

Confirmation of identity of the analyte is necessary when regulatory action is involved or the identity of the analyte is in question. Both chemical derivatization procedures and mass spectrometric analysis are most commonly employed. The chemical derivatization procedures are specific in converting the isolated toxin to a derivative, which exhibits different chromatographic and/or other physical properties from the parent compound. Identity is confirmed when the derivative from the analyte has the same characteristics as the derivative from the analytical standard. The most definitive method for confirming the identity of mycotoxins involves the application of mass spectrometry. The mass spectrum of compounds can be used to identify the analyte and elucidate unknown chemical structures.

In summary, analytical data of high creditability is obtained through the use of properly evaluated methods and the adherence to quality assurance principles by qualified analysts. The data can be used for science-based risk assessments to control mycotoxins in the food and feed supply.